sandro / bench_press Goto Github PK

View Code? Open in Web Editor NEW

43.0 4.0 8.0 186 KB

Abandoned. simple ruby benchmarking

Home Page: http://rubybenchmark.com

License: MIT License

Ruby 100.00%

bench_press's Introduction

Bench Press

Bench Press is a simple dsl around Ruby's Benchmark library.

Place the code you wish to benchmark inside of a measure block, run the bench_press command and you'll get a markdown report containing your system information and the realtime benchmark.

Additionally, benchmarks can be published to the Ruby Benchmark via bench_press --publish benchmark.rb command.

Creating a new benchmark

Use the --new flag to get started $ bench_press --new benchmark $ vi benchmark.rb

Example

# foo.rb
require 'bench_press'
extend BenchPress

base_string = ""
measure "string append" do
  base_string << "Hello World"
end

base_string = ""
measure "string +=" do
  base_string += "Hello World"
end

$ bench_press foo.rb

Foo
===
Date: August 05, 2010

System Information
------------------
    Operating System:    Mac OS X 10.6.4 (10F569)
    CPU:                 Intel Core 2 Duo 2.4 GHz
    Processor Count:     2
    Memory:              4 GB
    ruby 1.8.7 (2009-12-24 patchlevel 248) [i686-darwin10.2.0], MBARI 0x6770, Ruby Enterprise Edition 2010.01

"string append" is up to 71% faster over 1000 repetitions
---------------------------------------------------------

    string append    0.00270986557006836 secs    Fastest
    string +=        0.00948691368103027 secs    71% Slower

Details

The default number of repetitions is 1000 meaning each measure block is run 1000 times.

Each measure block is run in a forked subprocess in an attempt to isolate the memory usage per measurement. As of 0.3.0, the benchmark is run twice, the first run gets thrown away while the second run is added to the report.

Running the binary/examples locally

I use rubygems but this library is $LOAD_PATH friendly which means we need to set up our own load path when playing locally.

Try sourcing the .dev file

$ source .dev

Note on Patches/Pull Requests

Fork the project.
Make your feature addition or bug fix.
Add tests for it. This is important so I don't break it in a future version unintentionally.
Commit, do not mess with rakefile, version, or history. (if you want to have your own version, that is fine but bump version in a commit by itself I can ignore when I pull)
Send me a pull request. Bonus points for topic branches.

Copyright

bench_press's People

Contributors

Stargazers

Watchers

Forkers

groxx kaiwren rbxbx kbrock manveru dmeremyanin guniorobot nishu

bench_press's Issues

Cannot publish

I'm on Archlinux with Ruby 1.9.1.
After applying the patch in my fork: http://github.com/manveru/bench_press/commit/e0ff0dbbe752787cef5de6de2157f7f19bd4d578
I get this error when trying to publish.
Haven't digged much further, might be an issue with Facter instead.

/home/manveru/github/manveru/bench_press/lib/bench_press/system_information.rb:34:in `digest': can't convert nil into String (TypeError)
  from /home/manveru/github/manveru/bench_press/lib/bench_press/system_information.rb:34:in `hexdigest'
  from /home/manveru/github/manveru/bench_press/lib/bench_press/system_information.rb:34:in `crypted_identifier'
  from /home/manveru/github/manveru/bench_press/lib/bench_press/report.rb:42:in `to_hash'
  from /home/manveru/github/manveru/bench_press/lib/bench_press/ruby_benchmark.rb:9:in `publish'
  from /home/manveru/github/manveru/bench_press/lib/bench_press/cli.rb:92:in `publish'
  from /home/manveru/github/manveru/bench_press/lib/bench_press/cli.rb:68:in `perform_bench_press'
  from /home/manveru/github/manveru/bench_press/lib/bench_press/cli.rb:55:in `run'
  from /home/manveru/github/manveru/bench_press/bin/bench_press:4:in `<main>'
zsh: exit 1     ~/github/manveru/bench_press/bin/bench_press -p sum.rb

Add benchmark collaboration

This was a great start.

Have you thought of allowing people to submit their benchmark scripts?
And providing a service that records different results?

Maybe store something like:
github or twitter id, date, environment, and result.

Environment could possibly be: machine specs, benchmark, operating system, ruby version, gem version (not sure about this)

Thanks,
Keenan

String.any? doesn't work on ruby 1.9+

cli.rb, on line 89
if email && !email.nil?
Crash on ruby 1.9+
Using
if email && !email.nil?
Works good

Error when trying to run test

[13:20:13 miker@optisol-006 ~/optisol/src/rails/benchmarks] $ bench_press --publish tr_vs_gsub.rb
/usr/local/lib/ruby/gems/1.8/gems/bench_press-0.3.0/lib/bench_press/report.rb:59:in heading': undefined methodname' for nil:NilClass (NoMethodError)
from /usr/local/lib/ruby/gems/1.8/gems/bench_press-0.3.0/lib/bench_press/report.rb:63:in runnable_heading' from /usr/local/lib/ruby/gems/1.8/gems/bench_press-0.3.0/lib/bench_press/report.rb:20:into_s'
from /usr/local/lib/ruby/gems/1.8/gems/bench_press-0.3.0/lib/bench_press.rb:95:in puts' from /usr/local/lib/ruby/gems/1.8/gems/bench_press-0.3.0/lib/bench_press.rb:95 from /usr/local/bin/bench_press:19 /usr/local/lib/ruby/gems/1.8/gems/bench_press-0.3.0/lib/bench_press.rb:43:indate': undefined method parse' for Time:Class (NoMethodError) from /usr/local/share/optisol/src/rails/benchmarks/tr_vs_gsub.rb:8 from /usr/local/lib/ruby/gems/1.8/gems/bench_press-0.3.0/lib/bench_press/cli.rb:77:inload'
from /usr/local/lib/ruby/gems/1.8/gems/bench_press-0.3.0/lib/bench_press/cli.rb:77:in measurable' from /usr/local/lib/ruby/gems/1.8/gems/bench_press-0.3.0/lib/bench_press/cli.rb:66:inperform_bench_press'
from /usr/local/lib/ruby/gems/1.8/gems/bench_press-0.3.0/lib/bench_press/cli.rb:55:in run' from /usr/local/lib/ruby/gems/1.8/gems/bench_press-0.3.0/bin/bench_press:4 from /usr/local/bin/bench_press:19:inload'
from /usr/local/bin/bench_press:19

formatted_number broken below 1000

Hi, your current implementation of formatted_number does not work for numbers less than 1000. For example:

formatted_number(999)
=> nil

I see that your implementation borrows heavily from Rails:
http://api.rubyonrails.org/classes/ActionView/Helpers/NumberHelper.html#method-i-number_with_delimiter

The problem? gsub! can return nil if no substitutions are performed:

"Performs the substitutions of String#gsub in place, returning str, or nil if no substitutions were performed."
http://www.ruby-doc.org/core-1.9.3/String.html#method-i-gsub-21

The fix? Change your gsub! to gsub. Gotta love the one character fixes.

Ruby Interp. Should be Identified on main page

I like the web site! Newly found out about it. I think one important metric to skimming benchmarks is what version the interpreter is. I noticed the same test run twice on two different interps getting the opposite result:

Test August 2, 2010
"explicit return" is up to 35% faster over 100000 repetitions
Return August 2, 2010
"implicit return" is up to 12% faster over 100000 repetitions